Voltage-dependent Na+ channels are thought to sense membrane potential with fixed charges located within the membrane's electrical field. Measurement of open probability (Po) as a function of membrane potential gives a quantitative indication of the number of such charges that move through the field in opening the channel. We have used single-channel recording to measure skeletal muscle Na+ channel open probability at its most negative extreme, where channels may open as seldom as once per minute. To prevent fast inactivation from masking the voltage dependence of Po, we have generated a clone of the rat skeletal muscle Na+ channel that is lacking in fast inactivation (IFM1303QQQ). Using this mutant channel expressed in Xenopus oocytes, and the extra resolution afforded by single-channel analysis, we have extended the resolution of the hyperpolarized tail of the Po curve by four orders of magnitude. We show that previous measurements, which indicated a minimum of six effective gating charges, may have been made in a range of Po values that had not yet arrived at its limiting slope. In our preparation, a minimum of 12 charges must function in the activation gating of the channel. Our results will require reevaluation of kinetic models based on six charges, and they have major implications for the interpretation of S4 mutagenesis studies and structure/function models of the Na+ channel.